high-growth firms: not so vital after...

1

High-growth firms: Not so vital after all?

Sven-Olov Daunfeldt, Daniel Halvarsson, and Oana Mihaescu

Abstract: High-growth firms have received considerable interest recently since they create most of the new jobs in the economy. The purpose of our paper is to investigate the characteristics of high-growth firms prior to their growth period, and whether these characteristics differ across industries. Using data on a large sample of limited liability firms in Sweden for the period 2007-2010, we find that high-growth firms do not have the characteristics that we typically associate with successful firms. On the contrary, our results indicate that high-growth firms have low profits and a weak financial position. This might explain why studies have found that high-growth firms are seldom capable of sustaining their high growth rates in subsequent periods, and thus question policies that are targeted towards these companies. Keywords: Entrepreneurship; Firm growth; Gazelles; High-growth firms; High-impact firms; Innovation JEL-codes: L11; L25

2

1. Introduction

Studies have shown that a small number of high-growth firms (HGFs) are very important for job creation (Henrekson and Johansson, 2010; Coad et al., 2014a). Nesta (2009), for example, found that the 6 percent fastest growing firms in the United Kingdom created almost 50 percent of all jobs during 2002-2008. These firms were given the name “the vital 6 percent” to highlight their remarkable importance for job creation.1

HGFs’ ability to create job opportunities has attracted increasing attention from policymakers (Daunfeldt et al., 2015). Support for high-growth SMEs is, for example, stated as a political objective in the Europe 2020 strategy of the European Commission (European Commission, 2010). Some researchers also support the idea of targeting potential HGFs (e.g., Shane, 2009; Mason and Brown, 2013), arguing that policy should be redirected towards firms with growth aspirations instead of start-up firms with low survival rates.

The share of fast-growing innovative firms is emphasized by the European Commission (2010) as an important indicator to measure whether policies targeted towards potential HGFs are successful. However, the assumption that HGFs are overrepresented in high-tech industries seems to have little empirical support (Henrekson and Johansson, 2010). On the contrary, Daunfeldt et al. (2015) found that HGFs in Sweden were less common in R&D-intensive industries and overrepresented in knowledge-intensive service sectors. This points towards a “knowledge problem” embedded in the political initiative to promote the growth of HGFs, suggesting that we need more research on what actually characterizes the rapidly growing firms in the economy.

Previous studies have indicated that young (Reichstein et al., 2010; Barba Navaretti et al., 2014) and small firms (Birch, 1979; Almus, 2010; Goedhuys and Sleuwaegen, 2010) are more likely to be characterized by fast growth than older and larger firms. However, recent studies have argued that no systematic relationship exists between firm growth and firm size once age is controlled for (Haltiwanger et al., 2013; Lawless, 2014). The importance of firm age for explaining high-growth events is also highlighted by Daunfeldt et al. (2014), who identified HGFs in nine different ways and found that the common denominator regardless of definition was their relatively young age. There is also some evidence indicating that HGFs do not grow through acquisitions, but are more likely to enter into alliances with other firms (Mohr et al., 2014). HGFs also seem more likely to engage in export behavior than non-HGFs (Hölzl and Friesenbichler, 2007).

However, most previous studies on HGFs have not controlled for the profitability and the financial strength of the firms before they entered their period of fast growth. This is troublesome, considering that Penrose (1959) had long ago emphasized the importance of profits for achieving long-term growth. As stated by Brännback et al. (2009, p. 71):

1 Similar results have been shown by other researchers as well. Storey (1994), for example, found that the 4 percent fastest growing firms in the UK contributed to 50 percent of all jobs, and Daunfeldt et al. (2013) indicated that the 6 percent fastest growing firms generated 42 percent of all jobs in Sweden during 2005-2008.

3

“Being profitable clearly seems to be a far more productive, and in the long run, better approach to being a star firm.” Davidson et al. (2009) also found that the ability to grow in subsequent periods is positively associated with the firm’s profitability. It was found that, if initial growth coincides with high profitability, firms are more likely to display growth in future periods also.

Previous studies on the characteristics of HGFs have in most cases also been based on economy-wide data or data from selective industries, such as the manufacturing industry (Coad, 2009). This is unfortunate since we know that there are large differences across industries that might influence the likelihood of observing high-growth events (Audretsch et al., 2004). The manufacturing industry, for example, is capital intensive and characterized by high sunk costs, which is often interpreted as a sign that small and young manufacturing firms might be forced to grow fast in order to survive. On the other hand, such scale economies are less likely to be important for firms in low-tech service industries, such as the accommodation and food services industry.

We contribute to the HGF literature by investigating whether the characteristics of HGFs differ across industries after controlling for both profits and financial strength. Our analysis is based on a comprehensive dataset, covering all limited liability firms in Sweden during 2007-2010. We find that HGFs are characterized by low profits and low solidity prior to their growth episodes. This finding is perplexing since these character-features are atypical compared to the ones we associate with firms that achieve long-term growth. Viewed in light of recent studies (Hölzl, 2014; Daunfeldt and Halvarsson, 2015), however, it might help to explain why HGFs are unlikely to repeat their initial high growth rates in coming periods.

Policies to promote HGFs are often targeted towards R&D-intensive industries (Daunfeldt et al., 2015), but we find no evidence of substantial industry differences regarding the characteristics of HGFs prior to their growth period. As a consequence, we find no support for the view that HGFs within certain industries are more suitable to target than firms in other industries when designing industry policy.

In the next section, we present a brief theoretical background on why HGFs are seen as important job creators. Data and descriptive statistics are presented in Section 3, and our empirical method is described in Section 4. Results for the full sample and for five selected industries are then presented in Section 5. The final section concludes the paper with a summary and discussion of our key findings.

2. Theoretical background

Which firms are important for the creation of new jobs and economic growth? This question has interested researchers and policymakers for a long time. The answer, however, has shifted markedly over the past 100 years.

As far back as 1911, Schumpeter (1934/1911) emphasized the importance of entrepreneurs and new firms for creating economic growth and prosperity. According to the young Schumpeter, the entrepreneur was considered the individual force that

4

introduced new ideas in the economy, often by establishing new firms. In his view, these young firms, most of which were also small, were crucial for economic development and growth because they challenged the incumbents with new technology. Incumbents that could not keep up with the progress of these small, innovative firms were eventually replaced – a process that Schumpeter popularly called “creative destruction.” The image is clear yet powerful, portraying young companies as the destroyers of old, inefficient technology.

The older Schumpeter (1942), on the other hand, emphasized the importance of scale economies for both production and research and development. In the first decades after World War II, it was considered self-evident that it was the large and, naturally, the older companies that created jobs and growth. At this time, economic policy was attuned to the economic theories of return-to-scale production as was research and development. New and small businesses were viewed as inefficient. Occasionally, they were even considered a waste of resources (Galbraith, 1956, 1967). As a result, economic policies were designed to target large industrial companies.

In a very influential report, however, David Birch (1979) came to question this view. In accordance with the prevailing view at that time, he found that large firms accounted for the majority of all new jobs. However, when observing these firms over time, he found that large firms lost jobs and were replaced by firms that had once been small but had grown big. Small firms thus created the majority of all jobs over time, while larger businesses reduced their number of employees. The important insight and contribution was that the perception that large enterprises were important for job creation was based on a static approach, while the importance of small firms takes precedence in a dynamic analysis.

Birch’s (1979) results were controversial and criticized in several studies (e.g., Davis et al., 1996; Haltiwanger and Krizan, 1999). Later studies, however, confirmed most of his initial results (Van Praag and Versloot, 2008), but with one important addition: most small firms were not growing. The new jobs were instead being created by a small number of HGFs (Birch and Medoff, 1994; Henrekson and Johansson, 2010). Storey (1994) found, for example, that 50 percent of the new jobs in the UK were created by the 4 percent fastest growing companies. In a recent study, entitled The Vital 6 per cent, Nesta (2009) showed that it was rather the 6 percent fastest growing companies in the UK that accounted for half of all new jobs in the economy.

The job creation ability of HGFs has led to suggestions that policymakers should devote more resources to supporting these companies, rather than investing in start-ups that normally have no ambitions to grow or cannot survive market competition (Shane, 2009; Mason and Brown, 2013). The European Commission, for example, states in its strategy documents that more efforts should be directed towards supporting the fast-growing small and medium-sized firms (European Commission, 2010).

The idea of supporting potential HGFs has, however, been criticized recently since the growth of HGFs does not seem to be sustained over time (Hölzl, 2014; Daunfeldt and Halvarsson, 2015) and because it seems to be difficult to predict which firms will be

5

characterized by high growth in the future (Storey, 1994; Hölzl, 2009). Another problem with the orientation towards HGFs is that it leads to policies that are focused on R&D-intensive industries (Daunfeldt et al., 2015). However, there is little evidence that rapidly growing firms are more common in R&D-intensive industries (Hölzl, 2009; Henrekson and Johansson, 2010). Daunfeldt et al. (2015), for example, present results indicating that HGFs are less common in R&D-intensive industries and rather overrepresented in the knowledge-intensive service industries.

3. Data and descriptive statistics

All limited liability firms in Sweden are required by law to submit annual reports to the Swedish Patent and Registration Office (PRV). We use data from PAR, a Swedish consulting firm, which gathered this information from PRV. The data include information on all figures in the annual reports, such as profits, number of employees, industry classification, and sales.

In the dataset, firms are classified into industries according to the European Union’s NACE standard. We use these industry classification codes to select data on surviving firms in eight different industries during 2007-2010 (Table 1). We have selected these industries to induce possible variation in the likelihood of receiving policy interventions and the degree of technological knowledge. While some industries, such as manufacturing, are frequently studied (Coad, 2009) and are of considerable interest among politicians, industries such as hospitality and retail have received less attention. Compared to manufacturing with relatively high R&D expenditures, hospitality and retail are often considered to be low-tech industries that provide jobs for low-qualified workers and are rarely the focus of policy interventions. A similar observation can be made about policies that are targeted towards potential HGFs, which are dominated by R&D-intensive sectors. The relative lack of interest in retail and hospitality can also be observed in previous literature, with very few studies that specifically investigate the characteristics of fast-growing firms within these industries.2 Our final dataset consists of 78,937 firms active during 2007-2010.

[Table 1 about here]

One inescapable obstacle when investigating the characteristics of HGFs is that there is no consensus definition or way of identifying these firms (Coad et al., 2014). It is therefore necessary to make decisions regarding the growth indicator, firm growth measure, length of growth period, and growth process (Delmar and Davidsson, 1998).

Two of the most common growth indicators in the literature are sales and number of employees (Delmar et al., 2003; Daunfeldt et al., 2014), which are known to be modestly correlated (Shepherd and Wiklund, 2009). The indicators represent different aspects of the production process even if results seem little influenced when choosing one or the other (Daunfeldt et al., 2014). While the number of employees is an input factor (often

2 A notable exception is Daunfeldt et al. (2013), who investigated firm growth within the Swedish retail and wholesale trade industries during 2000-2004 using a quantile regression model.

6

considered to be quasi-fixed), sales represents a firm’s gross output. Thus, in using employment growth, “firm growth” captures the rate of change of internal resources, whereas using sales growth reflects the product’s or service’s acceptance in the market (Delmar et al., 2003). We therefore choose to apply both of these growth indicators in the paper. More specifically, employment growth and sales growth are calculated by:

∆ ln �� = ln �� − ln��,

∆ ln �� = ln �� − ln��, (1)

where ∆ ln �� is the logarithmic change in sales from 2007 to 2010, and ∆ ln �� the logarithmic change in the number of employees during the same period.3 It is worth noting here that the logarithmic difference is one of the most frequently used measures in the firm growth literature (Coad, 2009), with the convenient property of being symmetric for positive and negative growth rates (Tornqvist et al., 1985). This means that real changes in either indicator give the same percentage change, whether it is positive or negative.

As stated in equation (1), we follow the previous literature on HGFs and consider changes over the course of 3 years, more specifically between 2007 and 2010 (Coad et al., 2014). By avoiding annual growth rates we can avoid a lot of idiosyncrasy that exists for more narrowly defined growth rates. However, as Bjuggren et al. (2013) remark, results do not seem particularly sensitive to which period is chosen.

Finally, it is well known that growth can be divided into organic growth or acquired growth. Most studies do not have access to data on mergers and acquisitions and must therefore rely on measures of total growth, that is, the sum of organic and acquired growth. This is a drawback since what is interesting to investigate are the characteristics of and mechanisms by which firms achieve high levels of growth by increasing output and enhancing sales (organic growth), and not by mergers and acquisitions – growth that is generated outside the firm. Fortunately, the PAR database includes information on mergers and acquisitions and we use this information to exclude firms that have been subject to a merger or acquisition. In contrast to most previous studies, we can thus focus on organic growth and also control whether the firm was subject to a merger or acquisition before the study period.

Previous studies have indicated that HGFs tend to be young and small, and we therefore control for both firm age (Ai2014) and firm size (Ri2006, Ei2006) in our regressions. Firm age is measured using information on the registered start year, and defined as the observation year minus the registered start year. Access to registered start year is rare (Coad et al., 2015), and in contrast to previous studies we therefore do not need to work with truncated or censored age data. Firm size is measured using either sales or number of employees in 2006, depending on which growth indicator we use.

3 Note that relative growth measures, such as the one we apply, tend to favor smaller firms, whereas absolute growth measures are biased towards larger firms (Delmar et al., 2003).

7

Returns on total assets (ROAi2006) and solidity (Si2006) are included as independent variables in order to investigate the profitability and financial strength of HGFs prior to their growth period. We use returns on total assets as our profit measure since it is not affected by the type of financing (Libby et al., 2011), and although multiple profitability measures have been used in the literature (Richard et al., 2009), ROA is the one that seems most commonly used (Davidsson et al., 2009; Steffens et al., 2009). Financial strength of the firms is measured by their solidity, that is, the percentage share of equity out of total capital in 2006. This shows the relative proportion of equity that is used to finance a firm’s assets and indicates the firm’s solvency in the long term. The higher the proportion of equity that finances the company, the higher and better the solidity and the lower the financial risk.

Contrary to previous studies, we are also able to control whether the firm has been subject to a merger and acquisition before the study period (Ci2006). As we want to control whether the initial conditions of the firms influence their subsequent growth rates, all of our control variables (except firm age, which is a monotonic transformation) are measured in 2006. Finally, we also include municipality-specific fixed effects (�� and ��) and industry-specific fixed effects4 (�� and ��) to control for regional and industry

time-invariant heterogeneity that might affect firm growth rates.

Descriptive statistics of the variables that we include in our empirical analysis are presented in Table 2.


The descriptive statistics indicate that the average firm in the sample is around 18 years old, has eight employees, 5.4 percent in returns on total assets, and a solidity of 34.7. Sales are on average 16,509,000 SEK (corresponds in 31 August 2015 to 1,737,000 EUR), while the median firm’s sales are 2,585,000 SEK (272,000 EUR). Note also that more observations are missing when we investigate employment growth. The reason is that some firms with positive sales do not have any employees, which means that they will be excluded when we take the log difference to calculate firm growth rates.

4. Empirical method

It is well known that firm growth tends to follow the “tent-shaped” Laplace distribution (Stanley et al., 1996; Bottazzi and Secchi, 2006; Bottazzi et al., 2011), with most firms not growing and a few firms growing very fast. The familiar tent-shaped distribution is also evident in our dataset for both employment growth and sales growth (Figure 1). This violates the standard least-squares assumption of normally distributed error terms, and means that OLS estimation becomes less attractive. It is of little interest to estimate the average effect when the average firm is characterized by very marginal

4 Industry-specific fixed effects are used only for estimations conducted on the aggregated dataset including all eight industries considered for analysis.

8

growth rates. Neither do we want to consider HGFs as outliers, as OLS would, since our focus is on investigating what determines the growth rates of these fast-growing firms.

[Figure 1 about here]

Median regression, which assumes the error terms to be Laplace distributed, becomes more suitable in our case. Following Fotopoulos and Louri (2004), Coad and Rao (2008), and Reichstein et al. (2010), we therefore estimate a quantile regression model to investigate what characterizes firms across the entire growth rate distribution, including the fastest growing firms.

The estimated equations can be written:

∆ �� = �� + �� ∗ �� + �� ∗ �� + �� ∗ �� + �� ∗ �� +�� ∗ �� + �� + �� + !��

∆ �� = �� + �� ∗ �� + �� ∗ �� + �� ∗ �� + �� ∗ �� +�� ∗ �� + �� + �� + !�� (2)

where �� is the age of firm i; �� is number of employees in 2006; �� is firm sales in 2006; �� is returns on total assets in 2006; �� is the solidity of firm i in 2006; and �� is an indicator variable that equals one if the company has been subject to a merger or acquisition in 2006. We also include industry-specific fixed effects5, �� and ��, to account for time-invariant differences across industries that might

influence firm growth rates, and municipality-specific fixed effects, �� and ��, to account for time-invariant heterogeneity across municipalities in Sweden;6 �� and �� are constants, and ��- �� and ��- �� are parameters to be estimated. Finally, !�� and !�� are random error terms.

5. What characterizes high-growth firms?

5.1 Results, all firms

The results when equation (2) is estimated for all firms are presented in Table 3 (sales growth) and Table 4 (employment growth). In order to evaluate the appropriateness of using OLS, we present results both from an OLS model and from a quantile regression model.

The OLS results indicate that older firms are characterized by faster sales growth than younger firms (Table 3). However, the quantile regression results reveal that this result is driven by firms with marginal growth rates and that sales growth is not significantly related to firm age for the fastest growing firms. This shows the importance of not relying on OLS when investigating determinants of firm growth rates. No statistically significant

5 Idem. 6 We have also estimated a model without municipality- and/or industry-specific (in the case of all firms) fixed effects, and most results remain qualitatively similar. The results are available in Appendices B and C.

9

relationship is observed between firm size and firm growth in the OLS results, whereas firm growth seems to be negatively related to firm size for firms in the 0.80 quantile.

Profitability and solidity are two firm-specific variables that have seldom been controlled for in previous firm growth studies, although some researchers have argued that they might be important determinants for firm growth rates (Davidsson et al., 2009; Steffens et al., 2009). The OLS results in Table 3 indicate that sales growth is positively related to initial profits for the average firm. However, sales HGFs are less likely to be profitable, suggesting that they tend to grow before achieving profits. Solidity is also negatively related to rapid growth, implying that sales HGFs start their growth period from a weak financial position. Finally, sales HGFs are less likely to have participated in a merger and acquisition prior to their growth period.

In Table 4, the corresponding results for growth in number of employees are presented. Firm age is now positively related to firm growth for the majority of the firms in growth rate distribution, but not for the fastest growing firms. According to the results, employment HGFs are younger than firms that are growing more slowly. Firm size, on the other hand, does not seem to influence employment growth for the fastest growing firms. Neither is it statistically significant when estimating an OLS model. This implies that firm age is a more important determinant of employment growth than firm size, supporting Haltiwanger et al.’s (2003) findings.

The results also indicate that employment HGFs are more likely to be characterized by low profitability and a low degree of solidity. Firms that are growing fast in terms of number of employees thus seem to have lower initial profits and less financial strength than firms that are growing more slowly.

Note finally that the results for the 0.5 quantile are difficult to interpret since in most of the firms in this quantile the number of employees does not change, which means that the variation in the dependent variable is very low.



5.2 Industry differences among high-growth firms

The analysis has so far been focused on the characteristics of HGFs for all eight industries presented in Table 1. In order to test whether the characteristics of HGFs differ across industries, we have also estimated equation (2) separately for the following five industries: (i) Manufacturing (NACE-code 25); (ii) Construction (NACE-codes 42 and 43); (iii) Retail (NACE-code 47); (iv) Hospitality (NACE-codes 55 and 56); and (v) Computer programming (NACE-codes 62 and 63). The results for the fastest growing firms are presented in Table 5 (sales growth) and Table 6 (employment growth), while the results for the other growth quantiles are presented in Appendix A.


10


The results in Tables 5 and 6 indicate that HGFs in most cases share the same

characteristics across industries. Only small differences across the industries under study can be observed, and in most cases they confirm the aggregated results presented in Tables 3 and 4. Thus, despite considerable industry differences regarding scale economies, capital intensity, and human capital, HGFs are characterized by low initial profits and have a weak financial position prior to the growth period. The only differences that we can observe are that firms in the computer industry, that is, more knowledge-intensive services, seem to be older and have taken part in a merger before the study period.

6. Conclusions

The purpose of this paper has been to analyze the characteristics of HGFs in Sweden during 2007-2010, and to investigate whether they differ for firms active in different industries. This question is of importance since policymakers have started to design policies that are targeted towards potential HGFs in R&D-intensive industries (Daunfeldt et al., 2015).

HGFs were found to be characterized by low profits prior to their growth period, which is troublesome since profits seem important in predicting future growth (Davidsson et al., 2009; Steffens et al., 2009). The lack of profits would be less of a problem if HGFs were financially strong, but we found that HGFs had also grown from a weak financial position. This implies that HGFs do not have the characteristics that we typically associate with firms that are able to become successful in the long run. We believe that our results might help to explain why recent studies (e.g., Hölzl, 2014; Daunfeldt and Halvarsson, 2015) have found that HGFs are “one-hit wonders,” unlikely to sustain their high growth rates in subsequent periods. Our study thus adds reason for concern regarding the efficacy of policies that are targeted towards HGFs.

Policies that are targeted towards potential HGFs have in general been focused on R&D-intensive industries. However, we did not find any large differences among HGFs belonging to five industries that are very different in terms of, for example, capital intensity, minimum efficient scale, and share of educated workers. Thus, our results do not seem to be driven by industry-specific differences.

Our results question the current fascination with HGFs, and suggest that policies targeted towards these firms are unlikely to be successful. Maybe policymakers should instead focus on improving the general conditions for firm growth. As noted by Bornhäll et al. (2015), many profitable firms might choose to grow if the conditions for firm growth become more favorable. This implies that politicians should try to remove growth barriers for all firms instead of trying to pick winners or design policies targeted towards those firms that have high historical growth rates. We believe, therefore, that there is a need for more research on the conditions for firm growth, and on the kinds of policies that can promote firm growth that is sustainable in the long run.

11

Acknowledgments

Research funding from the R&D Fund of the Swedish Tourism and Hospitality Industry (BFUF) is gratefully acknowledged. A special thanks to Karl Wennberg.

12

REFERENCES

Almus, M. (2010). What characterizes a fast-growing firm? Applied Economics, 34(12), 1497-1508. Audretsch, D. B., Klomp, L., Santarelli, E., and Thurik, A. R. (2004). Gibrat's law: are the services different?. Review of Industrial Organization, 24(3), 301-324. Barba Navaretti, G., Castellani, D., and Pieri, F. (2014). Age and firm growth: evidence from three European countries. Small Business Economics, 43(4), 823-837. Birch, D. L. (1979), The Job Generation Process. MIT Program on Neighborhood and Regional Change. Massachusetts Institute of Technology: Cambridge, MA. Birch, D. L., and Medoff, J. (1994). Gazelles, in: L. C. Solmon & A. R. Levenson (Eds), Labor Markets, Employment Policy and Job Creation, 159-167, Boulder, CO: Westview. Bjuggren, C. M., Daunfeldt, S. O., and Johansson, D. (2013). High-growth firms and family ownership. Journal of Small Business & Entrepreneurship, 26(4), 365-385. Bornhäll, A., Daunfeldt, S-O., and Rudholm, N. (2015). Sleeping gazelles: The unseen job creators?. In Corbett, A.C., Katz, J.A., and McKelvie, A. Entrepreneurial growth: Individual, firm, and region, Advances in Entrepreneurship, Firm Emergence, and

Growth, 17, 161-185. Bottazzi G., and Secchi A. (2006). Explaining the distribution of firm growth rates. Rand Journal of Economics, 37(2), 235-256. Bottazzi, G., Coad, A., Jacoby, N., & Secchi, A. (2011). Corporate growth and industrial dynamics: Evidence from French manufacturing. Applied Economics, 43(1), 103-116. Brännback, M., Carsrud, A. L., and Kiviluoto, N. (2014). Understanding the Myth of

High Growth Firms. New York: Springer. Coad, A. (2009). The Growth of Firms: A Survey of Theories and Empirical Evidence. Edward Elgar, Cheltenham: United Kingdom. Coad, A., Daunfeldt, S-O., Hölzl, W., Johansson, D., and Nightingale, P. (2014a). High-growth firms: Introduction to the special issue. Industrial and Corporate Change, 23(1), 91-112. Coad, A., Daunfeldt, S-O., and Halvarsson, D. (2015). Bursting into life: Firm growth and growth persistence by age. HUI Working Paper 112, HUI Research: Stockholm. Coad, A., and Rao, R. (2008). Innovation and firm growth in high-tech sectors: A quantile regression approach. Research Policy, 37(4), 633–648.

13

Daunfeldt, S-O., Lang, Å., Macuchova, Z., and Rudholm, N. (2013). Firm growth in the retail and wholesale trade sectors: Evidence from Sweden”, The Service Industries

Journal, 33, 1193-1205. Daunfeldt, S-O., Elert, N., and Johansson, D. (2014). Economic contribution of high-growth firms: Do policy implications depend on the choice of growth indicator? Journal of Industry, Competition and Trade, 14(3), 337-365. Daunfeldt, S-O., Elert, N., and Johansson, D. (2015). Are high-growth firms overrepresented in high-tech industries? Forthcoming Industrial and Corporate

Change. Daunfeldt, S-O., and Halvarsson, D. (2015). Are High-Growth Firms One-Hit Wonders? Evidence from Sweden. Small Business Economics, 44(2), 361-383. Davidsson, P., Steffens., P., and Fitzsimmons, J. (2009). Growing profitable or growing from profits? Putting the horse in front of the cart? Journal of Business Venturing, 24(4), 388-406. Davis, S. J., Haltiwanger, J., & Schuh S. (1996). Small business and job creation: dissecting the myth and reassessing the facts. Small Business Economics, 8(4), 297-315. Delmar, F., Davidsson., P., and Gartner, W.B. (2003). Arriving at the high-growth firm. Journal of Business Venturing, 18(2), 189–216. Delmar, F., and Davidsson, P. (1998). A taxonomy of high-growth firms. 399-413 in: Reynolds, P.D., et al, (eds.), Frontiers of Entrepreneurship Research. Babson College, Wellesley, MA. European Commission (2010). Europe 2020: A strategy for smart, sustainable and inclusive growth: Communication from the Commission. Research report. Fotopoulos, G., and Louri, H. (2004). Firm growth and FDI: Are multinationals stimulating local industrial development?. Journal of Industry, Competition and Trade, 4(3), 163–189. Galbraith, J.K. (1956). American capitalism: The theory of countervailing power. Boston, MA: Houghton Mifflin. Galbraith, J.K. (1967). The new industrial state. Princeton, NJ: Princeton University Press. Goedhuys, M. and Sleuwaegen, L. (2010). High-growth entrepreneurial firms in Africa: a quantile regression approach. Small Business Economics, 34(1), 31-51. Haltiwanger, J., and Krizan, C. J. (1999). Small business and job creation in the United States: The role of new and young businesses. In Are small firms important? Their role

and impact (s. 79-97). Springer US.

14

Haltiwanger J., Jarmin R.S., and Miranda, J. (2013). Who creates jobs? Small versus Large versus Young. Review of Economics and Statistics, 95(2), 347–361. Henrekson M., and Johansson, D. (2010). Gazelles as job creators: a survey and interpretation of the evidence. Small Business Economics, 35(2), 227-244. Hölzl, W., and Friesenbichler, K. (2007). Are gazelles more innovative than other firms?. WIFO, Austria (mimeo). Hölzl, W. (2009). Is the R&D behaviour of fast-growing SMEs different? Evidence from CIS III data for 16 countries. Small Business Economics, 33(1), 59-75. Hölzl, W. (2014). Persistence, survival, and growth: a closer look at 20 years of fast-growing firms in Austria. Industrial and Corporate Change, 23(1), 199-231. Lawless, M., (2014). Age or size? Contributions to job creation. Small Business

Economics, 42(4), 815–830. Libby, R., Libby, P., and Short, D.G. (2011). Financial Accounting, 7 ed, McGraw-Hill/Irwin: New York. Mason, C., and Brown, R. (2013). Creating good public policy to support high-growth firms. Small Business Economics, 40(2), 211-225. Mohr, V., Garnsey, E., and Theyel, G. (2014). The role of alliances in the early development of high-growth firms. Industrial and Corporate Change, 23(1), 233-259. NESTA (2009), The Vital 6 Per Cent. NESTA: London, UK. Penrose, E. (1959). A Theory of the Growth of the Firm. 4th edition. Oxford University Press: New York. Reichstein, T., Dahl, M.S., Ebersberger, B., and Jensen, M.B. (2010). The devil dwells in the tails: A quantile regression approach to firm growth. Journal of Evolutionary

Economics, 20(2), 219-231. Richard, P.J., Devinney,T.T., Yip, G., and Johnson, G. (2009). Measuring organizational performance: Towards best methodological practice. Journal of

Management, 34(3), 718-804. Schumpeter, J. (1911/1934). The Theory of Economic Development. Harvard University Press, Cambridge. Schumpeter, J. (1942). Capitalism, Socialism and Democracy. Harper, New York. Shane, S. (2009). Why encouraging more people to become entrepreneurs is bad public policy. Small Business Economics, 33(2), 141-149.

15

Shepherd, D., and Wiklund, J. (2009). Are we comparing apples with apples or apples with oranges? Appropriateness of knowledge accumulation across growth studies. Entrepreneurship Theory and Practice, 33(1), 105-123 Stanley, M.H.R., Amaral, L.A.N., Buldyrev, S.V., Havlin, S., Leschhorn, H., Maass, P., Salinger, M.A., and Stanley, H.E. (1996). Scaling behavior in the growth of companies. Nature, 379(6568), 804-806. Storey, D. (1994). Understanding the Small Business Sector. Routledge, London. Tornqvist L., Vartia P., and Vartia Y.O. (1985). How should relative changes be measured? American Statistician 39(1), 43-46. Van Praag, M., and Versloot, P. (2008). The economic benefits and costs of entrepreneurship: A review of the research. Foundations and Trends in

Entrepreneurship, 4(2), 65-154.

16

Table 1. Industries and number of firms included in the analysis

NACE Industry Frequency Percent

43 Specialized construction activities 24,961 31.62 47 Retail trade, except motor vehicles and 24,767 31.38 Motorcycles 62 Computer programming 11,472 14.53 56 Food and beverage serving activities 8,654 10.96 25 Manufacture of fabricated metal products, 6,373 8.07 except machinery and equipment 55 Accommodation 1,157 1.47 42 Civil engineering 826 1.05 63 Information service activities 727 0.92 Total 78,937 100

Table 2. Descriptive statistics Variable Obs. Mean Median Std. Dev. Min Max

∆�� 50,729 0.0689 -0.0244 1.0881 -10.350 11.397 ∆�� 45,169 -0.0171 0 0.468 -4.736 5.193 �� 78,937 18.293 15 13.073 4 116 �� 60,292 16,509 2,585 250,401 0 33,700,000 �� 59,387 8.274 3 79.233 0 10,856 �� 58,596 5.417 6.2 44.999 -999 999 �� 58,591 34.674 37 59.877 -999 622 �� 78,937 0.212 0 0.409 0 1

17

Table 3. Estimation results, sales growth (∆"#$), 2007-2010, municipality and industry fixed effects. All firms. OLS and quantile regressions. Standard errors in parentheses.

Variable OLS Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0035*** 0.0060*** 0.0043*** 0.0024*** 0.0005 -0.0008 (0.0004) (0.000) (0.000) (0.000) (0.000) (0.001) *#&''+ -0.0001 0.00005 0.0000*** -0.0001** -0.0001*** -0.0002 (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) ",%#&''+ 0.00005*** 0.0002*** 0.0001*** 0.000001 -0.0001** -0.0002** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0003*** 0.0001 0.0000 -0.0001 -0.0006*** -0.0011*** (0.0001) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0306*** -0.0273* -0.0205*** -0.0196*** -0.0387*** -0.0842*** (0.0114) (0.014) (0.007) (0.005) (0.012) (0.029) constant 0.0754 -0.7426*** -0.3712*** 0.0634* 0.5650*** 1.1260*** (0.0877) (0.090) (0.081) (0.036) (0.053) (0.091) Observations 45,507 46,507 46,507 46,507 46,507 46,507 R-squared 0.0103 0.0025 0.0035 0.0052 0.0013 0.0011 Pseudo R-squared

0.0039 a 0.0448 0.0268 0.0106 0.0156 0.0247

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared.

Table 4. Estimation results, employment growth (∆*#$), 2007-2010, municipality and industry fixed effects. All firms. OLS and quantile regressions. Standard errors in parentheses.

Variable OLS Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.00267*** 0.0062*** 0.0039*** 0.0000 0.0009*** -0.0008* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) "#&''+ 0.0000 0.0000 0.0000 -0.0000 0.0000*** -0.0000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00003*** 0.000006 -0.00007*** -0.0000 -0.0001*** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0002*** 0.0002 0.0006*** 0.0000 -0.0012*** -0.0014*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0089* -0.0274*** -0.0441*** -0.0000 0.0474*** -0.0002 (0.005) (0.010) (0.009) (0.001) (0.009) (0.020) constant -0.0244 -0.6316*** -0.3166*** -0.0000 0.2415*** 0.5381*** (0.039) (0.032) (0.059) (0.004) (0.028) (0.049) Observations 42,133 42,133 42,133 42,133 42,133 42,133 R-squared 0.0176 0.0105 0.0098 0.00009 0.0067 0.0021 Pseudo R-squared

0.0105 a 0.0521 0.0335 0.0000 0.0246 0.0262

Standard errors in parentheses. *** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared.

18

Table 5. Estimation results sales-HGFs (0.90 percentile), per industry, 2007-2010. Standard errors in parentheses.

Industry

Variable Manufacturing Construction Retail Hospitality Computer

%#&'() -0.0005 0.00006 -0.0024** -0.0012 0.0077* (0.001) (0.001) (0.001) (0.003) (0.004) *#&''+ -0.0004 -0.0003*** -0.00002*** -0.0003*** -0.0005*** (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.0003* -0.0002 -0.0001 -0.0002** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0001 -0.0004 -0.0015*** -0.0008*** 0.00013 (0.001) (0.001) (0.001) (0.000) (0.001) .#&''+ -0.0710 -0.0723** 0.0047 0.0924 -0.0831 (0.052) (0.035) (0.050) (0.077) (0.086) constant 1.0965*** 0.6443*** 1.0441*** 1.9468*** 0.8408*** (0.191) (0.152) (0.109) (0.107) (0.285) Observations 4,728 16,744 14,528 4,619 5,888 R-squared 0.0273 0.0056 0.0102 0.0309 0.0176 Pseudo R-squared 0.1190 0.0319 0.0543 0.1340 0.0786

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. Table 6. Estimation results employment-HGFs (0.90 percentile), per industry, 2007-2010. Standard errors in parentheses.

Industry

Variable Manufacturing Construction Retail Hospitality Computer

%#&'() -0.0011* -0.0005 -0.00001 0.0005 0.0013 (0.001) (0.001) (0.001) (0.002) (0.001) "#&''+ -0.000** 0.000*** -0.000 -0.000* -0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.0002** -0.0002*** -0.0002*** -0.0001*** -0.0001* (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0003 -0.0007* -0.0013*** -0.0011** -0.0012*** (0.001) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0265 -0.0296 -0.0095 0.0686 0.1672** (0.023) (0.026) (0.018) (0.048) (0.066) constant 0.1860*** 0.3414*** 0.4085*** 0.4410*** 0.7504*** (0.062) (0.081) (0.056) (0.047) (0.121) Observations 4,412 15,403 13,187 4,125 5,006 R-squared 0.0328 0.0070 0.0053 0.0324 0.0108 Pseudo R-squared 0.1480 0.0457 0.0487 0.1210 0.1010

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level.

19

Figure 1. Frequency distributions, change in total sales and number of employees, 2007-2010.

Logarithmic change in total sales Logarithmic change in number of employees

20

APPENDIX A. Estimation results, per industry, with municipality

fixed effects Table A1. Estimation results, sales growth (∆"#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-code 25, Manufacturing. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0019* 0.0036** 0.0028*** 0.0014** -0.0001 -0.0005 (0.001) (0.002) (0.001) (0.001) (0.001) (0.001) *#&''+ -0.0002 -0.00002 -0.0001 0.00007 -0.0002 -0.0004 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.000008 0.0003** 0.0001 -0.00001 -0.0001 -0.0003* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.0007** 0.0005 0.0001 0.00029 0.0003 -0.0001 (0.000) (0.001) (0.001) (0.000) (0.001) (0.001) .#&''+ 0.0025 0.04057 0.02947 0.0149 -0.0393 -0.07097 (0.028) (0.035) (0.023) (0.018) (0.031) (0.052) constant 0.2131 -0.5512** -0.1948** 0.0891 0.5379*** 1.0965*** (0.2167) (0.222) (0.095) (0.065) (0.137) (0.191) Observations 4,728 4,728 4,728 4,728 4,728 4,728 R-squared 0.0565 0.0111 0.01402 0.0209 0.0258 0.0273 Pseudo R-squared

-0.0059 a 0.0934 0.0523 0.0398 0.0593 0.119

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table A2. Estimation results, employment growth (∆*#$), 2007-2010, municipality fixed effects. OLS and quantile regression, NACE-code 25, Manufacturing. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0013*** 0.0027*** 0.0013*** 0.000002 -0.0004 -0.0011* (0.000) (0.001) (0.000) (0.000) (0.001) (0.001) "#&''+ 0.0000 0.0000 0.0000 0.0000** -0.0000 -0.0000** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0000 -0.00002 -0.00003 -0.0000 -0.0001 -0.0002** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0003 0.0004 0.0004 -0.0000 -0.0009** -0.0003 (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0031 -0.0022 -0.0165 0.0257*** 0.0040 -0.0265 (0.0137) (0.020) (0.013) (0.007) (0.019) (0.023) constant -0.0054 -0.2708** -0.1474 -0.0001 0.1346*** 0.1860*** (0.1082) (0.137) (0.129) (0.022) (0.040) (0.062) Observations 4,412 4,412 4,412 4,412 4,412 4,412 R-squared 0.0716 0.0159 0.0230 0.0323 0.0370 0.0328 Pseudo R-squared

0.0058 a 0.122 0.0805 0.0226 0.0922 0.148


21

Table A3. Estimation results, sales growth (∆"#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 42 and 43, Construction. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0050*** 0.0085*** 0.0067*** 0.0033*** 0.0016** 0.00006 (0.001) (0.001) (0.000) (0.000) (0.001) (0.001) *#&''+ -0.0001 0.0001 0.00004 -0.00005 -0.0002*** -0.0003*** (0.0001) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0001*** 0.0002** 0.00009*** -0.000006 -0.00005 -0.0002 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0006*** -0.0003 -0.0002 -0.0002 -0.0005* -0.0004 (0.0002) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0623*** -0.0434* -0.0217 -0.0290*** -0.0518** -0.0723** (0.0186) (0.023) (0.013) (0.009) (0.022) (0.035) constant -0.0126 -0.7000*** -0.4522*** -0.0713 0.3307*** 0.6443*** (0.1106) (0.075) (0.125) (0.091) (0.065) (0.152) Observations 16,744 16,744 16,744 16,744 16,744 16,744 R-squared 0.0200 0.0075 0.0082 0.0089 0.0078 0.0056 Pseudo R-squared

0.0025 a 0.0416 0.0279 0.0124 0.0171 0.0319

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table A4. Estimation results, employment growth (∆*#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 42 and 43, Construction. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0031*** 0.0078*** 0.0048*** -0.0000 0.0008** -0.0005 (0.000) (0.001) (0.000) (0.000) (0.000) (0.001) "#&''+ 0.0000 0.0000 0.0000* 0.0000 0.0000*** 0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.0001*** -0.0000 -0.0002*** 0.0000 -0.00004** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0001 0.0002 0.0011*** -0.0000 -0.0006** -0.0007* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0151 -0.01509 -0.0250* 0.0000 0.0341* -0.0296 (0.009) (0.018) (0.013) (0.002) (0.020) (0.026) constant -0.1455*** -0.7753*** -0.4992*** -0.0000 0.0387 0.3414*** (0.0534) (0.033) (0.104) (0.007) (0.025) (0.081) Observations 15,403 15,403 15,403 15,403 15,403 15,403 R-squared 0.0252 0.0126 0.0120 0.0010 0.0092 0.0070 Pseudo R-squared

0.0063 a 0.0660 0.0452 0.0007 0.0337 0.0457


22

Table A5. Estimation results, sales growth (∆"#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-code 47, Retail. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0025*** 0.0043*** 0.0030*** 0.0019*** 0.0004 -0.0024** (0.001) (0.000) (0.000) (0.000) (0.001) (0.001) *#&''+ -0.00003 0.00007*** 0.00002 -0.00008*** -0.00001* -0.00002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.00006** 0.0002*** 0.00008*** 0.000007 -0.0001** -0.0001 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.00056*** 0.0004 -0.00003 -0.0002 -0.0013*** -0.0015*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.02846 -0.0639*** -0.0434*** -0.0225*** -0.0306 0.0047 (0.021) (0.020) (0.010) (0.008) (0.019) (0.050) constant -0.0376 -1.1206*** -0.5492*** -0.1560** 0.3688** 1.0441*** (0.186) (0.160) (0.161) (0.076) (0.178) (0.109) Observations 14,528 14,528 14,528 14,528 14,528 14,528 R-squared 0.0229 0.0032 0.0039 0.0055 0.0065 0.0102 Pseudo R-squared

0.0028 a 0.0498 0.0232 0.0094 0.0189 0.0543

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table A6. Estimation results, employment growth (∆*#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-code 47, Retail. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0021*** 0.0044*** 0.0025*** 0.0009** -0.00002 (0.000) (0.001) (0.000) (0.000) (0.001) "#&''+ 0.0000 0.0000*** 0.0000* 0.0000*** -0.0000 (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00003** 0.00002 -0.00007*** -0.0001*** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.000219** 0.0004 0.0005*** -0.0010*** -0.0013*** (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0220** -0.0468** -0.0609*** 0.0330** -0.0095 (0.009) (0.018) (0.013) (0.013) (0.018) constant -0.0385 -0.4621*** -0.2746*** 0.2810*** 0.4085*** (0.076) (0.058) (0.055) (0.038) (0.056) Observations 13,187 13,187 13,187 13,187 13,187 R-squared 0.0253 0.0103 0.0104 0.0074 0.0053 Pseudo R-squared

0.0031 a 0.0642 0.0433 0.0355 0.0487


23

Table A7. Estimation results, sales growth (∆"#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 55 and 56, Hospitality. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0008 0.0034*** 0.0037*** 0.0008 -0.0011 -0.0012 (0.002) (0.001) (0.001) (0.001) (0.001) (0.003) *#&''+ 0.000002 0.0003*** 0.0002 0.00004 -0.00005 -0.0003*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00008* 0.00002 0.00001 -0.00002 -0.00008 -0.0002** (0.00005) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0006** -0.0003 -0.0002 -0.0002 -0.0004 -0.0008*** (0.0003) (0.001) (0.000) (0.000) (0.000) (0.000) .#&''+ 0.0833** 0.0274 0.0132 0.0205 0.0511 0.0924 (0.0407) (0.039) (0.024) (0.016) (0.034) (0.077) constant 0.1461 -3.7921*** -0.1864** -0.0325 1.5020*** 1.9468*** (0.4042) (0.148) (0.085) (0.091) (0.138) (0.107) Observations 4,619 4,619 4,619 4,619 4,619 4,619 R-squared 0.0633 0.0092 0.0142 0.0248 0.0316 0.0309 Pseudo R-squared

0.0033 a 0.107 0.0575 0.0273 0.0672 0.134

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table A8. Estimation results, employment growth (∆*#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 55 and 56, Hospitality. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0043*** 0.0077*** 0.0059*** -0.0000 0.0017 0.0005 (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) "#&''+ 0.0000002 0.0000*** 0.0000 0.0000 0.0000 -0.0000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00003 0.00001 -0.000007 -0.0000 -0.00009** -0.0001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0008*** -0.0006 -0.0005*** -0.0000 -0.0010** -0.0011** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ 0.0324 0.0539 0.0141 0.0000 0.0092 0.0686 (0.022) (0.038) (0.027) (0.008) (0.022) (0.048) constant -0.3324 -0.7697*** -0.7422*** -0.1335 0.4183*** 0.4410*** (0.2797) (0.068) (0.054) (0.116) (0.047) (0.047) Observations 4,125 4,125 4,125 4,125 4,125 4,125 R-squared 0.0763 0.0266 0.0399 0.0317 0.0381 0.0324 Pseudo R-squared

0.0113 a 0.131 0.0939 0.0259 0.0858 0.121


24

Table A9. Estimation results, sales growth (∆"#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 62 and 63, Computer. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0069*** 0.0087*** 0.0071*** 0.0065*** 0.0065 0.0077* (0.002) (0.003) (0.002) (0.001) (0.004) (0.004) *#&''+ -0.0002 0.0002*** 0.00007 -0.0001 -0.0003*** -0.0005*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0001*** 0.0003*** 0.0002* 0.0001 -0.0001** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.00007 -0.0004 0.0005 0.0006 0.0005 0.0001 (0.000) (0.000) (0.001) (0.001) (0.001) (0.001) .#&''+ -0.0429 -0.0141 -0.0585 -0.0732*** -0.0684 -0.0831 (0.038) (0.053) (0.041) (0.023) (0.052) (0.086) constant -0.3995 -1.443229*** -0.8442** -0.2435 0.1102 0.8408*** (0.315) (0.311) (0.386) (0.164) (0.134) (0.285) Observations 5,888 5,888 5,888 5,888 5,888 5,888 R-squared 0.0472 0.0138 0.0200 0.0298 0.0241 0.0176 Pseudo R-squared

0.0032 a 0.0830 0.0505 0.0296 0.0520 0.0786

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table A10. Estimation results, employment growth (∆*#$), 2007-2010, municipality fixed effects. OLS and quantile regression. NACE-codes 62 and 63, Computer. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0047*** 0.0086*** 0.0060*** 0.0000* 0.000002 0.0013 (0.001) (0.001) (0.001) (0.000) (0.000) (0.001) "#&''+ -0.0000 0.0000 -0.0000 -0.0000*** -0.0000*** -0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00004** 0.00002 -0.00002 0.0000** -0.0000 -0.0001* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.0002 0.0004** 0.0010*** -0.0000 -0.000002 -0.0012*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0016 -0.0735*** -0.1717*** -0.000002 0.2140*** 0.1672** (0.016) (0.028) (0.039) (0.000) (0.031) (0.066) constant -0.0655 -0.8070*** -0.2297*** 0.000004 0.0002 0.7504*** (0.119) (0.142) (0.087) (0.000) (0.017) (0.121) Observations 5,006 5,006 5,006 5,006 5,006 5,006 R-squared 0.052 0.0246 0.0227 0.0189 0.0116 0.0108 Pseudo R-squared

0.0016 a 0.1120 0.1070 0.0139 0.0699 0.1010


25

APPENDIX B. Estimation results, per industry, without municipality fixed

effects

Table B1. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. OLS and quantile regressions. NACE-code 25, Manufacturing. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0022** 0.0040*** 0.0025*** 0.0018*** 0.00002 0.0018 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) *#&''+ -0.0002 -0.00004 0.00002 0.00007 -0.0002* -0.0006*** (0.000) (0.000) (0.001) (0.000) (0.000) (0.000) ",%#&''+ 0.000002 0.0004** 0.00009 -0.000009 -0.0002*** -0.0003*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.0009*** 0.0008 0.0004 0.0003 0.0003 0.0008 (0.000) (0.002) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0080 0.0370 0.0243 0.0196 -0.0214 -0.1330*** (0.027) (0.038) (0.026) (0.015) (0.023) (0.051) constant 0.1258*** -0.6240*** -0.3048*** 0.0721*** 0.5331*** 0.8814*** (0.028) (0.109) (0.030) (0.018) (0.030) (0.059) Observations 4,728 4,728 4,728 4,728 4,728 4,728 R-squared 0.0035 0.0017 0.0022 0.0018 0.0000 0.0002 Pseudo R-squared

0.0024 a 0.0110 0.0048 0.0020 0.0021 0.0070

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table B2. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-code 25, Manufacturing. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0013*** 0.0035*** 0.0019*** 0.000001 -0.0003 -0.0012 (0.000) (0.001) (0.000) (0.000) (0.000) (0.002) "#&''+ 0.00000 0.0000*** 0.0000 0.0000 -0.0000*** -0.0000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0000 0.0000 -0.0001** -0.0000 -0.0001* -0.0003*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0002 0.0014** 0.0014*** -0.0000 -0.0005*** -0.0002 (0.000) (0.001) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0080 0.01482 -0.0218 0.0508*** -0.0001 -0.1366** (0.013) (0.025) (0.018) (0.014) (0.017) (0.056) constant 0.0533*** -0.4667*** -0.2158*** -0.00003 0.3245*** 0.6566*** (0.014) (0.028) (0.025) (0.006) (0.018) (0.049) Observations 4,412 4,412 4,412 4,412 4,412 4,412 R-squared 0.0022 0.0003 0.0002 0.0000 0.0001 0.0000 Pseudo R-squared

0.0011 a 0.0142 0.0127 0.0026 0.0014 0.0128


26

Table B3. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. OLS and quantile regressions. NACE-codes 42 and 43, Construction. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0050*** 0.0090*** 0.0067*** 0.0033*** 0.0014** -0.0001 (0.001) (0.001) (0.000) (0.000) (0.001) (0.001) *#&''+ -0.0001 0.000* 0.000003 -0.00005** -0.0001*** -0.0002*** (0.0001356) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0001** 0.0002*** 0.00009*** -0.000004 -0.00006* -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0006*** -0.0003 -0.0001 -0.0003*** -0.0005** -0.0005 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0644*** -0.0522** -0.0296** -0.0217*** -0.0658*** -0.0795** (0.0183) (0.021) (0.013) (0.008) (0.018) (0.036) constant -0.0281* -0.8063*** -0.5127*** -0.0936*** 0.37182*** 0.8027*** (0.0167) (0.024) (0.014) (0.009) (0.019) (0.037) Observations 16,744 16,744 16,744 16,744 16,744 16,744 R-squared 0.0049 0.0041 0.0041 0.0043 0.0016 0.00002 Pseudo R-squared

0.0046 a 0.0188 0.0140 0.0038 0.0015 0.0019

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table B4. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-codes 42 and 43, Construction. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0031*** 0.0087*** 0.0055*** 0.0000 0.0015*** -0.0009 (0.0003) (0.001) (0.000) (0.000) (0.000) (0.001) "#&''+ 0.0000 0.0000 -0.0000 0.0000 0.0000*** 0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00007*** -0.0001* -0.0002*** 0.0000 -0.0001** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0001 0.0002 0.0012*** 0.0000 -0.0019*** -0.0007* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0160* -0.0227 -0.0307** 0.0000 0.0402*** -0.0477** (0.0090) (0.036) (0.012) (0.002) (0.013) (0.022) constant -0.0800*** -0.6875*** -0.4034*** 0.00000 0.1569*** 0.4990*** (0.008) (0.050) (0.020) (0.003) (0.023) (0.037) Observations 15,403 15,403 15,403 15,403 15,403 15,403 R-squared 0.0083 0.0078 0.0057 0.0011 0.0002 Pseudo R-squared

0.0080 a 0.0295 0.0219 0 0.0075 0.0023


27

Table B5. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-code 47, Retail. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0025*** 0.0053*** 0.0032*** 0.001945*** -0.0003 -0.0040** (0.001) (0.001) (0.000) (0.000) (0.001) (0.002) *#&''+ -0.000 0.000*** -0.00002 -0.000088*** -0.00001 -0.00004*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.0001** 0.0002*** 0.0001 0.000006 -0.00007*** -0.0002** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0005*** 0.0002 0.00007 -0.000193 -0.0013*** -0.0018*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0359* -0.0970*** -0.0451*** -0.022175*** -0.0335* -0.0460 (0.020) (0.025) (0.009) (0.007) (0.020) (0.069) constant 0.1077*** -0.5630*** -0.3108*** -0.045791*** 0.3819*** 0.9921*** (0.018) (0.019) (0.012) (0.008) (0.020) (0.052) Observations 14,528 14,528 14,528 14,528 14,528 14,528 R-squared 0.0018 0.0012 0.0011 0.0014 0.0000 0.0001 Pseudo R-squared

0.0015 a 0.0175 0.0087 0.00277 0.0028 0.0058

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table B6. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-code 47, Retail. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0027*** 0.0052*** 0.0032*** 0.0000 0.0009* -0.0005 (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) "#&''+ 0.0000 0.0000*** 0.0000 0.0000 0.0000*** 0.0000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.0000** 0.00002 -0.00008*** 0.0000 -0.0001*** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0002** 0.0004 0.0006*** 0.0000 -0.0017*** -0.0015*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0249*** -0.0760*** -0.0706*** 0.0000 0.0395*** 0.0059 (0.009) (0.015) (0.012) (0.001) (0.013) (0.020) constant -0.0278*** -0.5064*** -0.2495*** 0.0000 0.1882*** 0.4910*** (0.008) (0.024) (0.014) (0.002) (0.018) (0.021) Observations 13,187 13,187 13,187 13,187 13,187 13,187 R-squared 0.0063 0.0042 0.0038 0.000772 0.0003 Pseudo R-squared

0.0059 a 0.0190 0.0161 0 0.00955 0.0054


28

Table B7. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-codes 55 and 56, Hospitality. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0014 0.0037*** 0.0033*** 0.0006 -0.0014 0.0041 (0.002) (0.001) (0.001) (0.001) (0.001) (0.004) *#&''+ -0.00008 0.0004*** 0.0001** 0.00003 -0.000054** -0.0004*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00010** -0.00003 0.000007 -0.00001 -0.00014 -0.0003 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0005* -0.0002 -0.00005 -0.0001 -0.00064 -0.0015 (0.000) (0.001) (0.000) (0.000) (0.001) (0.003) .#&''+ 0.0759* -0.0285 0.0242 0.0119 0.03664 0.1220 (0.0394) (0.053) (0.020) (0.013) (0.034) (0.159) constant .0062 -0.6927*** -0.4317*** -0.0961*** 0.2893*** 0.7844*** (0.0355) (0.037) (0.018) (0.013) (0.034) (0.100) Observations 4,619 4,619 4,619 4,619 4,619 4,619 R-squared 0.0030 0.0005 0.0004 0.0026 0.00214 0.0029 Pseudo R-squared

0.0020 a 0.0022 0.0035 0.0005 0.0019 0.0049

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table B8. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-codes 55 and 56, Hospitality. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0046*** 0.0082*** 0.0063*** 0.0000 0.0028* 0.0024 (0.001) (0.001) (0.001) (0.001) (0.002) (0.004) "#&''+ 0.0000002 0.0000*** 0.0000*** 0.0000 0.0000 -0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00005* 0.000001 -0.000007 0.0000 -0.00008 -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0008*** -0.0005*** -0.0004*** 0.0000 -0.0016* -0.0018 (0.000) (0.000) (0.000) (0.000) (0.001) (0.002) .#&''+ 0.0365* 0.0801** 0.034295 0.0000 0.0284 0.0828 (0.021) (0.034) (0.024) (0.007) (0.033) (0.106) constant -0.1036*** -0.7542*** -0.4593*** 0.0000 0.1726*** 0.5412*** (0.01863) (0.022) (0.021) (0.014) (0.039) (0.080) Observations 4,125 4,125 4,125 4,125 4,125 4,125 R-squared 0.0139 0.0110 0.0115 0.0098 0.0089 Pseudo R-squared

0.0127 a 0.0225 0.0154 0.0000 0.0091 0.0074


29

Table B9. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-codes 62 and 63, Computer. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0073*** 0.0096*** 0.0087*** 0.0068*** 0.0082* 0.0061 (0.002) (0.003) (0.001) (0.001) (0.004) (0.005) *#&''+ -0.0002 0.0001** 0.0001** -0.00008** -0.0003*** -0.0005*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.00009*** 0.0002*** 0.0002*** 0.00005 -0.00008*** -0.0001* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.0001 0.0002 0.0008 0.0005 0.0006*** 0.0009 (0.000) (0.001) (0.001) (0.001) (0.000) (0.001) .#&''+ -0.0692* -0.0084 -0.0684** -0.0897*** -0.0886* -0.1674** (0.037) (0.071) (0.030) (0.020) (0.053) (0.085) constant -0.0209 -1.2092*** -0.7138*** -0.1443*** 0.5184*** 1.2829*** (0.041) (0.066) (0.032) (0.035) (0.081) (0.098) Observations 5,888 5,888 5,888 5,888 5,888 5,888 R-squared 0.0044 0.0034 0.0037 0.0037 0.0010 0.0004 Pseudo R-squared

0.0036 a 0.0123 0.0103 0.0047 0.0034 0.0037

*** significant at the 1% level; ** significant at the 5% level; * significant at the 10% level. a Adjusted R-squared. Table B10. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. OLS and quantile regression. NACE-codes 62 and 63, Computer. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0050*** 0.01076*** 0.0063*** 0.0000 0.000004 0.0024 (0.001) (0.001) (0.001) (0.000) (0.000) (0.004) "#&''+ -0.0000 -0.0000*** -0.0000* -0.0000*** -0.0000*** -0.0000*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.00004** 0.000009 -0.00005* 0.0000*** -0.0000 -0.0001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ 0.0002 0.0005** 0.0020*** -0.0000 -0.000003 -0.0015* (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0049 -0.0646*** -0.2034*** -0.000001 0.2512*** 0.2066*** (0.015) (0.023) (0.031) (0.000) (0.022) (0.054) constant -0.1231*** -0.8100*** -0.3685*** 0.000008*** 0.0003 0.4880*** (0.017) (0.014) (0.047) (0.000) (0.003) (0.059) Observations 5,006 5,006 5,006 5,006 5,006 5,006 R-squared 0.0099 0.0073 0.0018 0.0001 0.00003 0.0007 Pseudo R-squared

0.0089 a 0.0225 0.0421 0.0002 0.0299 0.0190


30

APPENDIX C. Estimation results, all firms, without municipality or

industry fixed effects Table C1. Estimation results, sales growth (∆"#$), 2007-2010, no fixed effects. All firms. OLS and quantile regressions. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0038*** 0.0083*** 0.0056*** 0.0027*** 0.0003 -0.0033*** (0.0004) (0.000) (0.000) (0.000) (0.000) (0.001) *#&''+ -0.00008 0.00004*** 0.000004 -0.00007 -0.0001*** -0.0002** (0.0001) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ 0.00005*** 0.0002*** 0.0001*** -0.000001 -0.00008*** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0003*** -0.0001 -0.0001 -0.00004 -0.0002 -0.0007 (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) .#&''+ -0.0270** -0.068*** -0.0301*** -0.0144*** -0.0182 -0.0461 (0.011) (0.016) (0.007) (0.005) (0.012) (0.030) constant 0.0348*** -0.770*** -0.4486*** -0.0732*** 0.4046*** 0.9955*** (0.010) (0.018) (0.012) (0.006) (0.015) (0.040) Observations 46,507 46,507 46,507 46,507 46,507 46,507 R-squared 0.0024 0.0021 0.0021 0.0021 0.0000 0.0006 Pseudo R-squared

0.0023a 0.0174 0.0113 0.0028 0.0007 0.0024


Table C2. Estimation results, employment growth (∆*#$), 2007-2010, no fixed effects. All firms. OLS and quantile regressions. Standard errors in parentheses.

Variable OLS

regression

Quantile regression

0.10 0.20 0.50 0.80 0.90

%#&'() 0.0029*** 0.0071*** 0.0046*** 0.0000 0.0017*** -0.0012*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) "#&''+ 0.0000 0.0000*** 0.0000 0.0000 0.0000*** 0.0000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ",%#&''+ -0.0000*** -0.000008 -0.00007*** 0.00000 -0.0001*** -0.0002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) -#&''+ -0.0002*** 0.0004* 0.0009*** 0.0000 -0.0019*** -0.0013*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) .#&''+ -0.0072 -0.0466*** -0.0519*** 0.0000 0.0613*** 0.0044 (0.005) (0.011) (0.007) (0.001) (0.009) (0.012) constant -0.0570*** -0.6138*** -0.3423*** 0.0000 0.1775*** 0.5391*** (0.005) (0.021) (0.010) (0.002) (0.015) (0.014) Observations 42,133 42,133 42,133 42,133 42,133 42,133 R-squared 0.0076 0.0057 0.0038 - 0.0016 0.0001 Pseudo R-squared

0.0074a 0.0260 0.0177 0.0000 0.0099 0.0068


high-growth firms: not so vital after...

Documents